1 /*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Kenneth Almquist. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#)parser.c 8.7 (Berkeley) 5/16/95 37 * $FreeBSD: src/bin/sh/parser.c,v 1.111 2011/05/20 16:03:36 jilles Exp $ 38 */ 39 40 #include <stdio.h> 41 #include <stdlib.h> 42 #include <unistd.h> 43 44 #include "shell.h" 45 #include "parser.h" 46 #include "nodes.h" 47 #include "expand.h" /* defines rmescapes() */ 48 #include "syntax.h" 49 #include "options.h" 50 #include "input.h" 51 #include "output.h" 52 #include "var.h" 53 #include "error.h" 54 #include "memalloc.h" 55 #include "mystring.h" 56 #include "alias.h" 57 #include "show.h" 58 #include "eval.h" 59 #include "exec.h" /* to check for special builtins */ 60 #ifndef NO_HISTORY 61 #include "myhistedit.h" 62 #endif 63 64 /* 65 * Shell command parser. 66 */ 67 68 #define EOFMARKLEN 79 69 #define PROMPTLEN 128 70 71 /* values of checkkwd variable */ 72 #define CHKALIAS 0x1 73 #define CHKKWD 0x2 74 #define CHKNL 0x4 75 76 /* values returned by readtoken */ 77 #include "token.h" 78 79 80 81 struct heredoc { 82 struct heredoc *next; /* next here document in list */ 83 union node *here; /* redirection node */ 84 char *eofmark; /* string indicating end of input */ 85 int striptabs; /* if set, strip leading tabs */ 86 }; 87 88 struct parser_temp { 89 struct parser_temp *next; 90 void *data; 91 }; 92 93 94 static struct heredoc *heredoclist; /* list of here documents to read */ 95 static int doprompt; /* if set, prompt the user */ 96 static int needprompt; /* true if interactive and at start of line */ 97 static int lasttoken; /* last token read */ 98 MKINIT int tokpushback; /* last token pushed back */ 99 static char *wordtext; /* text of last word returned by readtoken */ 100 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 101 static struct nodelist *backquotelist; 102 static union node *redirnode; 103 static struct heredoc *heredoc; 104 static int quoteflag; /* set if (part of) last token was quoted */ 105 static int startlinno; /* line # where last token started */ 106 static int funclinno; /* line # where the current function started */ 107 static struct parser_temp *parser_temp; 108 109 110 static union node *list(int, int); 111 static union node *andor(void); 112 static union node *pipeline(void); 113 static union node *command(void); 114 static union node *simplecmd(union node **, union node *); 115 static union node *makename(void); 116 static void parsefname(void); 117 static void parseheredoc(void); 118 static int peektoken(void); 119 static int readtoken(void); 120 static int xxreadtoken(void); 121 static int readtoken1(int, char const *, char *, int); 122 static int noexpand(char *); 123 static void synexpect(int) __dead2; 124 static void synerror(const char *) __dead2; 125 static void setprompt(int); 126 127 128 static void * 129 parser_temp_alloc(size_t len) 130 { 131 struct parser_temp *t; 132 133 INTOFF; 134 t = ckmalloc(sizeof(*t)); 135 t->data = NULL; 136 t->next = parser_temp; 137 parser_temp = t; 138 t->data = ckmalloc(len); 139 INTON; 140 return t->data; 141 } 142 143 144 static void * 145 parser_temp_realloc(void *ptr, size_t len) 146 { 147 struct parser_temp *t; 148 149 INTOFF; 150 t = parser_temp; 151 if (ptr != t->data) 152 error("bug: parser_temp_realloc misused"); 153 t->data = ckrealloc(t->data, len); 154 INTON; 155 return t->data; 156 } 157 158 159 static void 160 parser_temp_free_upto(void *ptr) 161 { 162 struct parser_temp *t; 163 int done = 0; 164 165 INTOFF; 166 while (parser_temp != NULL && !done) { 167 t = parser_temp; 168 parser_temp = t->next; 169 done = t->data == ptr; 170 ckfree(t->data); 171 ckfree(t); 172 } 173 INTON; 174 if (!done) 175 error("bug: parser_temp_free_upto misused"); 176 } 177 178 179 static void 180 parser_temp_free_all(void) 181 { 182 struct parser_temp *t; 183 184 INTOFF; 185 while (parser_temp != NULL) { 186 t = parser_temp; 187 parser_temp = t->next; 188 ckfree(t->data); 189 ckfree(t); 190 } 191 INTON; 192 } 193 194 195 /* 196 * Read and parse a command. Returns NEOF on end of file. (NULL is a 197 * valid parse tree indicating a blank line.) 198 */ 199 200 union node * 201 parsecmd(int interact) 202 { 203 int t; 204 205 /* This assumes the parser is not re-entered, 206 * which could happen if we add command substitution on PS1/PS2. 207 */ 208 parser_temp_free_all(); 209 heredoclist = NULL; 210 211 tokpushback = 0; 212 doprompt = interact; 213 if (doprompt) 214 setprompt(1); 215 else 216 setprompt(0); 217 needprompt = 0; 218 t = readtoken(); 219 if (t == TEOF) 220 return NEOF; 221 if (t == TNL) 222 return NULL; 223 tokpushback++; 224 return list(1, 1); 225 } 226 227 228 static union node * 229 list(int nlflag, int erflag) 230 { 231 union node *ntop, *n1, *n2, *n3; 232 int tok; 233 234 checkkwd = CHKNL | CHKKWD | CHKALIAS; 235 if (!nlflag && !erflag && tokendlist[peektoken()]) 236 return NULL; 237 ntop = n1 = NULL; 238 for (;;) { 239 n2 = andor(); 240 tok = readtoken(); 241 if (tok == TBACKGND) { 242 if (n2->type == NCMD || n2->type == NPIPE) { 243 n2->ncmd.backgnd = 1; 244 } else if (n2->type == NREDIR) { 245 n2->type = NBACKGND; 246 } else { 247 n3 = (union node *)stalloc(sizeof (struct nredir)); 248 n3->type = NBACKGND; 249 n3->nredir.n = n2; 250 n3->nredir.redirect = NULL; 251 n2 = n3; 252 } 253 } 254 if (ntop == NULL) 255 ntop = n2; 256 else if (n1 == NULL) { 257 n1 = (union node *)stalloc(sizeof (struct nbinary)); 258 n1->type = NSEMI; 259 n1->nbinary.ch1 = ntop; 260 n1->nbinary.ch2 = n2; 261 ntop = n1; 262 } 263 else { 264 n3 = (union node *)stalloc(sizeof (struct nbinary)); 265 n3->type = NSEMI; 266 n3->nbinary.ch1 = n1->nbinary.ch2; 267 n3->nbinary.ch2 = n2; 268 n1->nbinary.ch2 = n3; 269 n1 = n3; 270 } 271 switch (tok) { 272 case TBACKGND: 273 case TSEMI: 274 tok = readtoken(); 275 /* FALLTHROUGH */ 276 case TNL: 277 if (tok == TNL) { 278 parseheredoc(); 279 if (nlflag) 280 return ntop; 281 } else if (tok == TEOF && nlflag) { 282 parseheredoc(); 283 return ntop; 284 } else { 285 tokpushback++; 286 } 287 checkkwd = CHKNL | CHKKWD | CHKALIAS; 288 if (!nlflag && !erflag && tokendlist[peektoken()]) 289 return ntop; 290 break; 291 case TEOF: 292 if (heredoclist) 293 parseheredoc(); 294 else 295 pungetc(); /* push back EOF on input */ 296 return ntop; 297 default: 298 if (nlflag || erflag) 299 synexpect(-1); 300 tokpushback++; 301 return ntop; 302 } 303 } 304 } 305 306 307 308 static union node * 309 andor(void) 310 { 311 union node *n1, *n2, *n3; 312 int t; 313 314 n1 = pipeline(); 315 for (;;) { 316 if ((t = readtoken()) == TAND) { 317 t = NAND; 318 } else if (t == TOR) { 319 t = NOR; 320 } else { 321 tokpushback++; 322 return n1; 323 } 324 n2 = pipeline(); 325 n3 = (union node *)stalloc(sizeof (struct nbinary)); 326 n3->type = t; 327 n3->nbinary.ch1 = n1; 328 n3->nbinary.ch2 = n2; 329 n1 = n3; 330 } 331 } 332 333 334 335 static union node * 336 pipeline(void) 337 { 338 union node *n1, *n2, *pipenode; 339 struct nodelist *lp, *prev; 340 int negate, t; 341 342 negate = 0; 343 checkkwd = CHKNL | CHKKWD | CHKALIAS; 344 TRACE(("pipeline: entered\n")); 345 while (readtoken() == TNOT) 346 negate = !negate; 347 tokpushback++; 348 n1 = command(); 349 if (readtoken() == TPIPE) { 350 pipenode = (union node *)stalloc(sizeof (struct npipe)); 351 pipenode->type = NPIPE; 352 pipenode->npipe.backgnd = 0; 353 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 354 pipenode->npipe.cmdlist = lp; 355 lp->n = n1; 356 do { 357 prev = lp; 358 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 359 checkkwd = CHKNL | CHKKWD | CHKALIAS; 360 t = readtoken(); 361 tokpushback++; 362 if (t == TNOT) 363 lp->n = pipeline(); 364 else 365 lp->n = command(); 366 prev->next = lp; 367 } while (readtoken() == TPIPE); 368 lp->next = NULL; 369 n1 = pipenode; 370 } 371 tokpushback++; 372 if (negate) { 373 n2 = (union node *)stalloc(sizeof (struct nnot)); 374 n2->type = NNOT; 375 n2->nnot.com = n1; 376 return n2; 377 } else 378 return n1; 379 } 380 381 382 383 static union node * 384 command(void) 385 { 386 union node *n1, *n2; 387 union node *ap, **app; 388 union node *cp, **cpp; 389 union node *redir, **rpp; 390 int t; 391 int is_subshell; 392 393 checkkwd = CHKNL | CHKKWD | CHKALIAS; 394 is_subshell = 0; 395 redir = NULL; 396 n1 = NULL; 397 rpp = &redir; 398 399 /* Check for redirection which may precede command */ 400 while (readtoken() == TREDIR) { 401 *rpp = n2 = redirnode; 402 rpp = &n2->nfile.next; 403 parsefname(); 404 } 405 tokpushback++; 406 407 switch (readtoken()) { 408 case TIF: 409 n1 = (union node *)stalloc(sizeof (struct nif)); 410 n1->type = NIF; 411 if ((n1->nif.test = list(0, 0)) == NULL) 412 synexpect(-1); 413 if (readtoken() != TTHEN) 414 synexpect(TTHEN); 415 n1->nif.ifpart = list(0, 0); 416 n2 = n1; 417 while (readtoken() == TELIF) { 418 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 419 n2 = n2->nif.elsepart; 420 n2->type = NIF; 421 if ((n2->nif.test = list(0, 0)) == NULL) 422 synexpect(-1); 423 if (readtoken() != TTHEN) 424 synexpect(TTHEN); 425 n2->nif.ifpart = list(0, 0); 426 } 427 if (lasttoken == TELSE) 428 n2->nif.elsepart = list(0, 0); 429 else { 430 n2->nif.elsepart = NULL; 431 tokpushback++; 432 } 433 if (readtoken() != TFI) 434 synexpect(TFI); 435 checkkwd = CHKKWD | CHKALIAS; 436 break; 437 case TWHILE: 438 case TUNTIL: { 439 int got; 440 n1 = (union node *)stalloc(sizeof (struct nbinary)); 441 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 442 if ((n1->nbinary.ch1 = list(0, 0)) == NULL) 443 synexpect(-1); 444 if ((got=readtoken()) != TDO) { 445 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 446 synexpect(TDO); 447 } 448 n1->nbinary.ch2 = list(0, 0); 449 if (readtoken() != TDONE) 450 synexpect(TDONE); 451 checkkwd = CHKKWD | CHKALIAS; 452 break; 453 } 454 case TFOR: 455 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 456 synerror("Bad for loop variable"); 457 n1 = (union node *)stalloc(sizeof (struct nfor)); 458 n1->type = NFOR; 459 n1->nfor.var = wordtext; 460 while (readtoken() == TNL) 461 ; 462 if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) { 463 app = ≈ 464 while (readtoken() == TWORD) { 465 n2 = (union node *)stalloc(sizeof (struct narg)); 466 n2->type = NARG; 467 n2->narg.text = wordtext; 468 n2->narg.backquote = backquotelist; 469 *app = n2; 470 app = &n2->narg.next; 471 } 472 *app = NULL; 473 n1->nfor.args = ap; 474 if (lasttoken != TNL && lasttoken != TSEMI) 475 synexpect(-1); 476 } else { 477 static char argvars[5] = { 478 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0' 479 }; 480 n2 = (union node *)stalloc(sizeof (struct narg)); 481 n2->type = NARG; 482 n2->narg.text = argvars; 483 n2->narg.backquote = NULL; 484 n2->narg.next = NULL; 485 n1->nfor.args = n2; 486 /* 487 * Newline or semicolon here is optional (but note 488 * that the original Bourne shell only allowed NL). 489 */ 490 if (lasttoken != TNL && lasttoken != TSEMI) 491 tokpushback++; 492 } 493 checkkwd = CHKNL | CHKKWD | CHKALIAS; 494 if ((t = readtoken()) == TDO) 495 t = TDONE; 496 else if (t == TBEGIN) 497 t = TEND; 498 else 499 synexpect(-1); 500 n1->nfor.body = list(0, 0); 501 if (readtoken() != t) 502 synexpect(t); 503 checkkwd = CHKKWD | CHKALIAS; 504 break; 505 case TCASE: 506 n1 = (union node *)stalloc(sizeof (struct ncase)); 507 n1->type = NCASE; 508 if (readtoken() != TWORD) 509 synexpect(TWORD); 510 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 511 n2->type = NARG; 512 n2->narg.text = wordtext; 513 n2->narg.backquote = backquotelist; 514 n2->narg.next = NULL; 515 while (readtoken() == TNL); 516 if (lasttoken != TWORD || ! equal(wordtext, "in")) 517 synerror("expecting \"in\""); 518 cpp = &n1->ncase.cases; 519 checkkwd = CHKNL | CHKKWD, readtoken(); 520 while (lasttoken != TESAC) { 521 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 522 cp->type = NCLIST; 523 app = &cp->nclist.pattern; 524 if (lasttoken == TLP) 525 readtoken(); 526 for (;;) { 527 *app = ap = (union node *)stalloc(sizeof (struct narg)); 528 ap->type = NARG; 529 ap->narg.text = wordtext; 530 ap->narg.backquote = backquotelist; 531 checkkwd = CHKNL | CHKKWD; 532 if (readtoken() != TPIPE) 533 break; 534 app = &ap->narg.next; 535 readtoken(); 536 } 537 ap->narg.next = NULL; 538 if (lasttoken != TRP) 539 synexpect(TRP); 540 cp->nclist.body = list(0, 0); 541 542 checkkwd = CHKNL | CHKKWD | CHKALIAS; 543 if ((t = readtoken()) != TESAC) { 544 if (t != TENDCASE) 545 synexpect(TENDCASE); 546 else 547 checkkwd = CHKNL | CHKKWD, readtoken(); 548 } 549 cpp = &cp->nclist.next; 550 } 551 *cpp = NULL; 552 checkkwd = CHKKWD | CHKALIAS; 553 break; 554 case TLP: 555 n1 = (union node *)stalloc(sizeof (struct nredir)); 556 n1->type = NSUBSHELL; 557 n1->nredir.n = list(0, 0); 558 n1->nredir.redirect = NULL; 559 if (readtoken() != TRP) 560 synexpect(TRP); 561 checkkwd = CHKKWD | CHKALIAS; 562 is_subshell = 1; 563 break; 564 case TBEGIN: 565 n1 = list(0, 0); 566 if (readtoken() != TEND) 567 synexpect(TEND); 568 checkkwd = CHKKWD | CHKALIAS; 569 break; 570 /* Handle an empty command like other simple commands. */ 571 case TBACKGND: 572 case TSEMI: 573 case TAND: 574 case TOR: 575 /* 576 * An empty command before a ; doesn't make much sense, and 577 * should certainly be disallowed in the case of `if ;'. 578 */ 579 if (!redir) 580 synexpect(-1); 581 case TNL: 582 case TEOF: 583 case TWORD: 584 case TRP: 585 tokpushback++; 586 n1 = simplecmd(rpp, redir); 587 return n1; 588 default: 589 synexpect(-1); 590 } 591 592 /* Now check for redirection which may follow command */ 593 while (readtoken() == TREDIR) { 594 *rpp = n2 = redirnode; 595 rpp = &n2->nfile.next; 596 parsefname(); 597 } 598 tokpushback++; 599 *rpp = NULL; 600 if (redir) { 601 if (!is_subshell) { 602 n2 = (union node *)stalloc(sizeof (struct nredir)); 603 n2->type = NREDIR; 604 n2->nredir.n = n1; 605 n1 = n2; 606 } 607 n1->nredir.redirect = redir; 608 } 609 610 return n1; 611 } 612 613 614 static union node * 615 simplecmd(union node **rpp, union node *redir) 616 { 617 union node *args, **app; 618 union node **orig_rpp = rpp; 619 union node *n = NULL; 620 int special; 621 622 /* If we don't have any redirections already, then we must reset */ 623 /* rpp to be the address of the local redir variable. */ 624 if (redir == 0) 625 rpp = &redir; 626 627 args = NULL; 628 app = &args; 629 /* 630 * We save the incoming value, because we need this for shell 631 * functions. There can not be a redirect or an argument between 632 * the function name and the open parenthesis. 633 */ 634 orig_rpp = rpp; 635 636 for (;;) { 637 if (readtoken() == TWORD) { 638 n = (union node *)stalloc(sizeof (struct narg)); 639 n->type = NARG; 640 n->narg.text = wordtext; 641 n->narg.backquote = backquotelist; 642 *app = n; 643 app = &n->narg.next; 644 } else if (lasttoken == TREDIR) { 645 *rpp = n = redirnode; 646 rpp = &n->nfile.next; 647 parsefname(); /* read name of redirection file */ 648 } else if (lasttoken == TLP && app == &args->narg.next 649 && rpp == orig_rpp) { 650 /* We have a function */ 651 if (readtoken() != TRP) 652 synexpect(TRP); 653 funclinno = plinno; 654 /* 655 * - Require plain text. 656 * - Functions with '/' cannot be called. 657 * - Reject name=(). 658 * - Reject ksh extended glob patterns. 659 */ 660 if (!noexpand(n->narg.text) || quoteflag || 661 strchr(n->narg.text, '/') || 662 strchr("!%*+-=?@}~", 663 n->narg.text[strlen(n->narg.text) - 1])) 664 synerror("Bad function name"); 665 rmescapes(n->narg.text); 666 if (find_builtin(n->narg.text, &special) >= 0 && 667 special) 668 synerror("Cannot override a special builtin with a function"); 669 n->type = NDEFUN; 670 n->narg.next = command(); 671 funclinno = 0; 672 return n; 673 } else { 674 tokpushback++; 675 break; 676 } 677 } 678 *app = NULL; 679 *rpp = NULL; 680 n = (union node *)stalloc(sizeof (struct ncmd)); 681 n->type = NCMD; 682 n->ncmd.backgnd = 0; 683 n->ncmd.args = args; 684 n->ncmd.redirect = redir; 685 return n; 686 } 687 688 static union node * 689 makename(void) 690 { 691 union node *n; 692 693 n = (union node *)stalloc(sizeof (struct narg)); 694 n->type = NARG; 695 n->narg.next = NULL; 696 n->narg.text = wordtext; 697 n->narg.backquote = backquotelist; 698 return n; 699 } 700 701 void 702 fixredir(union node *n, const char *text, int err) 703 { 704 TRACE(("Fix redir %s %d\n", text, err)); 705 if (!err) 706 n->ndup.vname = NULL; 707 708 if (is_digit(text[0]) && text[1] == '\0') 709 n->ndup.dupfd = digit_val(text[0]); 710 else if (text[0] == '-' && text[1] == '\0') 711 n->ndup.dupfd = -1; 712 else { 713 714 if (err) 715 synerror("Bad fd number"); 716 else 717 n->ndup.vname = makename(); 718 } 719 } 720 721 722 static void 723 parsefname(void) 724 { 725 union node *n = redirnode; 726 727 if (readtoken() != TWORD) 728 synexpect(-1); 729 if (n->type == NHERE) { 730 struct heredoc *here = heredoc; 731 struct heredoc *p; 732 int i; 733 734 if (quoteflag == 0) 735 n->type = NXHERE; 736 TRACE(("Here document %d\n", n->type)); 737 if (here->striptabs) { 738 while (*wordtext == '\t') 739 wordtext++; 740 } 741 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 742 synerror("Illegal eof marker for << redirection"); 743 rmescapes(wordtext); 744 here->eofmark = wordtext; 745 here->next = NULL; 746 if (heredoclist == NULL) 747 heredoclist = here; 748 else { 749 for (p = heredoclist ; p->next ; p = p->next); 750 p->next = here; 751 } 752 } else if (n->type == NTOFD || n->type == NFROMFD) { 753 fixredir(n, wordtext, 0); 754 } else { 755 n->nfile.fname = makename(); 756 } 757 } 758 759 760 /* 761 * Input any here documents. 762 */ 763 764 static void 765 parseheredoc(void) 766 { 767 struct heredoc *here; 768 union node *n; 769 770 while (heredoclist) { 771 here = heredoclist; 772 heredoclist = here->next; 773 if (needprompt) { 774 setprompt(2); 775 needprompt = 0; 776 } 777 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 778 here->eofmark, here->striptabs); 779 n = (union node *)stalloc(sizeof (struct narg)); 780 n->narg.type = NARG; 781 n->narg.next = NULL; 782 n->narg.text = wordtext; 783 n->narg.backquote = backquotelist; 784 here->here->nhere.doc = n; 785 } 786 } 787 788 static int 789 peektoken(void) 790 { 791 int t; 792 793 t = readtoken(); 794 tokpushback++; 795 return (t); 796 } 797 798 static int 799 readtoken(void) 800 { 801 int t; 802 struct alias *ap; 803 #ifdef DEBUG 804 int alreadyseen = tokpushback; 805 #endif 806 807 top: 808 t = xxreadtoken(); 809 810 /* 811 * eat newlines 812 */ 813 if (checkkwd & CHKNL) { 814 while (t == TNL) { 815 parseheredoc(); 816 t = xxreadtoken(); 817 } 818 } 819 820 /* 821 * check for keywords and aliases 822 */ 823 if (t == TWORD && !quoteflag) 824 { 825 const char * const *pp; 826 827 if (checkkwd & CHKKWD) 828 for (pp = parsekwd; *pp; pp++) { 829 if (**pp == *wordtext && equal(*pp, wordtext)) 830 { 831 lasttoken = t = pp - parsekwd + KWDOFFSET; 832 TRACE(("keyword %s recognized\n", tokname[t])); 833 goto out; 834 } 835 } 836 if (checkkwd & CHKALIAS && 837 (ap = lookupalias(wordtext, 1)) != NULL) { 838 pushstring(ap->val, strlen(ap->val), ap); 839 goto top; 840 } 841 } 842 out: 843 if (t != TNOT) 844 checkkwd = 0; 845 846 #ifdef DEBUG 847 if (!alreadyseen) 848 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 849 else 850 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 851 #endif 852 return (t); 853 } 854 855 856 /* 857 * Read the next input token. 858 * If the token is a word, we set backquotelist to the list of cmds in 859 * backquotes. We set quoteflag to true if any part of the word was 860 * quoted. 861 * If the token is TREDIR, then we set redirnode to a structure containing 862 * the redirection. 863 * In all cases, the variable startlinno is set to the number of the line 864 * on which the token starts. 865 * 866 * [Change comment: here documents and internal procedures] 867 * [Readtoken shouldn't have any arguments. Perhaps we should make the 868 * word parsing code into a separate routine. In this case, readtoken 869 * doesn't need to have any internal procedures, but parseword does. 870 * We could also make parseoperator in essence the main routine, and 871 * have parseword (readtoken1?) handle both words and redirection.] 872 */ 873 874 #define RETURN(token) return lasttoken = token 875 876 static int 877 xxreadtoken(void) 878 { 879 int c; 880 881 if (tokpushback) { 882 tokpushback = 0; 883 return lasttoken; 884 } 885 if (needprompt) { 886 setprompt(2); 887 needprompt = 0; 888 } 889 startlinno = plinno; 890 for (;;) { /* until token or start of word found */ 891 c = pgetc_macro(); 892 switch (c) { 893 case ' ': case '\t': 894 continue; 895 case '#': 896 while ((c = pgetc()) != '\n' && c != PEOF); 897 pungetc(); 898 continue; 899 case '\\': 900 if (pgetc() == '\n') { 901 startlinno = ++plinno; 902 if (doprompt) 903 setprompt(2); 904 else 905 setprompt(0); 906 continue; 907 } 908 pungetc(); 909 goto breakloop; 910 case '\n': 911 plinno++; 912 needprompt = doprompt; 913 RETURN(TNL); 914 case PEOF: 915 RETURN(TEOF); 916 case '&': 917 if (pgetc() == '&') 918 RETURN(TAND); 919 pungetc(); 920 RETURN(TBACKGND); 921 case '|': 922 if (pgetc() == '|') 923 RETURN(TOR); 924 pungetc(); 925 RETURN(TPIPE); 926 case ';': 927 if (pgetc() == ';') 928 RETURN(TENDCASE); 929 pungetc(); 930 RETURN(TSEMI); 931 case '(': 932 RETURN(TLP); 933 case ')': 934 RETURN(TRP); 935 default: 936 goto breakloop; 937 } 938 } 939 breakloop: 940 return readtoken1(c, BASESYNTAX, NULL, 0); 941 #undef RETURN 942 } 943 944 945 #define MAXNEST_STATIC 8 946 struct tokenstate 947 { 948 const char *syntax; /* *SYNTAX */ 949 int parenlevel; /* levels of parentheses in arithmetic */ 950 enum tokenstate_category 951 { 952 TSTATE_TOP, 953 TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */ 954 TSTATE_VAR_NEW, /* other ${var...}, own dquote state */ 955 TSTATE_ARITH 956 } category; 957 }; 958 959 960 /* 961 * Called to parse command substitutions. 962 */ 963 964 static char * 965 parsebackq(char *out, struct nodelist **pbqlist, 966 int oldstyle, int dblquote, int quoted) 967 { 968 struct nodelist **nlpp; 969 union node *n; 970 char *volatile str; 971 struct jmploc jmploc; 972 struct jmploc *const savehandler = handler; 973 int savelen; 974 int saveprompt; 975 const int bq_startlinno = plinno; 976 char *volatile ostr = NULL; 977 struct parsefile *const savetopfile = getcurrentfile(); 978 struct heredoc *const saveheredoclist = heredoclist; 979 struct heredoc *here; 980 981 str = NULL; 982 if (setjmp(jmploc.loc)) { 983 popfilesupto(savetopfile); 984 if (str) 985 ckfree(str); 986 if (ostr) 987 ckfree(ostr); 988 heredoclist = saveheredoclist; 989 handler = savehandler; 990 if (exception == EXERROR) { 991 startlinno = bq_startlinno; 992 synerror("Error in command substitution"); 993 } 994 longjmp(handler->loc, 1); 995 } 996 INTOFF; 997 savelen = out - stackblock(); 998 if (savelen > 0) { 999 str = ckmalloc(savelen); 1000 memcpy(str, stackblock(), savelen); 1001 } 1002 handler = &jmploc; 1003 heredoclist = NULL; 1004 INTON; 1005 if (oldstyle) { 1006 /* 1007 * We must read until the closing backquote, giving special 1008 * treatment to some slashes, and then push the string and 1009 * reread it as input, interpreting it normally. 1010 */ 1011 char *oout; 1012 int c, olen; 1013 1014 STARTSTACKSTR(oout); 1015 for (;;) { 1016 if (needprompt) { 1017 setprompt(2); 1018 needprompt = 0; 1019 } 1020 CHECKSTRSPACE(2, oout); 1021 switch (c = pgetc()) { 1022 case '`': 1023 goto done; 1024 1025 case '\\': 1026 if ((c = pgetc()) == '\n') { 1027 plinno++; 1028 if (doprompt) 1029 setprompt(2); 1030 else 1031 setprompt(0); 1032 /* 1033 * If eating a newline, avoid putting 1034 * the newline into the new character 1035 * stream (via the USTPUTC after the 1036 * switch). 1037 */ 1038 continue; 1039 } 1040 if (c != '\\' && c != '`' && c != '$' 1041 && (!dblquote || c != '"')) 1042 USTPUTC('\\', oout); 1043 break; 1044 1045 case '\n': 1046 plinno++; 1047 needprompt = doprompt; 1048 break; 1049 1050 case PEOF: 1051 startlinno = plinno; 1052 synerror("EOF in backquote substitution"); 1053 break; 1054 1055 default: 1056 break; 1057 } 1058 USTPUTC(c, oout); 1059 } 1060 done: 1061 USTPUTC('\0', oout); 1062 olen = oout - stackblock(); 1063 INTOFF; 1064 ostr = ckmalloc(olen); 1065 memcpy(ostr, stackblock(), olen); 1066 setinputstring(ostr, 1); 1067 INTON; 1068 } 1069 nlpp = pbqlist; 1070 while (*nlpp) 1071 nlpp = &(*nlpp)->next; 1072 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1073 (*nlpp)->next = NULL; 1074 1075 if (oldstyle) { 1076 saveprompt = doprompt; 1077 doprompt = 0; 1078 } 1079 1080 n = list(0, oldstyle); 1081 1082 if (oldstyle) 1083 doprompt = saveprompt; 1084 else { 1085 if (readtoken() != TRP) 1086 synexpect(TRP); 1087 } 1088 1089 (*nlpp)->n = n; 1090 if (oldstyle) { 1091 /* 1092 * Start reading from old file again, ignoring any pushed back 1093 * tokens left from the backquote parsing 1094 */ 1095 popfile(); 1096 tokpushback = 0; 1097 } 1098 STARTSTACKSTR(out); 1099 CHECKSTRSPACE(savelen + 1, out); 1100 INTOFF; 1101 if (str) { 1102 memcpy(out, str, savelen); 1103 STADJUST(savelen, out); 1104 ckfree(str); 1105 str = NULL; 1106 } 1107 if (ostr) { 1108 ckfree(ostr); 1109 ostr = NULL; 1110 } 1111 here = saveheredoclist; 1112 if (here != NULL) { 1113 while (here->next != NULL) 1114 here = here->next; 1115 here->next = heredoclist; 1116 heredoclist = saveheredoclist; 1117 } 1118 handler = savehandler; 1119 INTON; 1120 if (quoted) 1121 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1122 else 1123 USTPUTC(CTLBACKQ, out); 1124 return out; 1125 } 1126 1127 1128 /* 1129 * Called to parse a backslash escape sequence inside $'...'. 1130 * The backslash has already been read. 1131 */ 1132 static char * 1133 readcstyleesc(char *out) 1134 { 1135 int c, v, i, n; 1136 1137 c = pgetc(); 1138 switch (c) { 1139 case '\0': 1140 synerror("Unterminated quoted string"); 1141 case '\n': 1142 plinno++; 1143 if (doprompt) 1144 setprompt(2); 1145 else 1146 setprompt(0); 1147 return out; 1148 case '\\': 1149 case '\'': 1150 case '"': 1151 v = c; 1152 break; 1153 case 'a': v = '\a'; break; 1154 case 'b': v = '\b'; break; 1155 case 'e': v = '\033'; break; 1156 case 'f': v = '\f'; break; 1157 case 'n': v = '\n'; break; 1158 case 'r': v = '\r'; break; 1159 case 't': v = '\t'; break; 1160 case 'v': v = '\v'; break; 1161 case 'x': 1162 v = 0; 1163 for (;;) { 1164 c = pgetc(); 1165 if (c >= '0' && c <= '9') 1166 v = (v << 4) + c - '0'; 1167 else if (c >= 'A' && c <= 'F') 1168 v = (v << 4) + c - 'A' + 10; 1169 else if (c >= 'a' && c <= 'f') 1170 v = (v << 4) + c - 'a' + 10; 1171 else 1172 break; 1173 } 1174 pungetc(); 1175 break; 1176 case '0': case '1': case '2': case '3': 1177 case '4': case '5': case '6': case '7': 1178 v = c - '0'; 1179 c = pgetc(); 1180 if (c >= '0' && c <= '7') { 1181 v <<= 3; 1182 v += c - '0'; 1183 c = pgetc(); 1184 if (c >= '0' && c <= '7') { 1185 v <<= 3; 1186 v += c - '0'; 1187 } else 1188 pungetc(); 1189 } else 1190 pungetc(); 1191 break; 1192 case 'c': 1193 c = pgetc(); 1194 if (c < 0x3f || c > 0x7a || c == 0x60) 1195 synerror("Bad escape sequence"); 1196 if (c == '\\' && pgetc() != '\\') 1197 synerror("Bad escape sequence"); 1198 if (c == '?') 1199 v = 127; 1200 else 1201 v = c & 0x1f; 1202 break; 1203 case 'u': 1204 case 'U': 1205 n = c == 'U' ? 8 : 4; 1206 v = 0; 1207 for (i = 0; i < n; i++) { 1208 c = pgetc(); 1209 if (c >= '0' && c <= '9') 1210 v = (v << 4) + c - '0'; 1211 else if (c >= 'A' && c <= 'F') 1212 v = (v << 4) + c - 'A' + 10; 1213 else if (c >= 'a' && c <= 'f') 1214 v = (v << 4) + c - 'a' + 10; 1215 else 1216 synerror("Bad escape sequence"); 1217 } 1218 if (v == 0 || (v >= 0xd800 && v <= 0xdfff)) 1219 synerror("Bad escape sequence"); 1220 /* We really need iconv here. */ 1221 if (initial_localeisutf8 && v > 127) { 1222 CHECKSTRSPACE(4, out); 1223 /* 1224 * We cannot use wctomb() as the locale may have 1225 * changed. 1226 */ 1227 if (v <= 0x7ff) { 1228 USTPUTC(0xc0 | v >> 6, out); 1229 USTPUTC(0x80 | (v & 0x3f), out); 1230 return out; 1231 } else if (v <= 0xffff) { 1232 USTPUTC(0xe0 | v >> 12, out); 1233 USTPUTC(0x80 | ((v >> 6) & 0x3f), out); 1234 USTPUTC(0x80 | (v & 0x3f), out); 1235 return out; 1236 } else if (v <= 0x10ffff) { 1237 USTPUTC(0xf0 | v >> 18, out); 1238 USTPUTC(0x80 | ((v >> 12) & 0x3f), out); 1239 USTPUTC(0x80 | ((v >> 6) & 0x3f), out); 1240 USTPUTC(0x80 | (v & 0x3f), out); 1241 return out; 1242 } 1243 } 1244 if (v > 127) 1245 v = '?'; 1246 break; 1247 default: 1248 synerror("Bad escape sequence"); 1249 } 1250 v = (char)v; 1251 /* 1252 * We can't handle NUL bytes. 1253 * POSIX says we should skip till the closing quote. 1254 */ 1255 if (v == '\0') { 1256 while ((c = pgetc()) != '\'') { 1257 if (c == '\\') 1258 c = pgetc(); 1259 if (c == PEOF) 1260 synerror("Unterminated quoted string"); 1261 } 1262 pungetc(); 1263 return out; 1264 } 1265 if (SQSYNTAX[v] == CCTL) 1266 USTPUTC(CTLESC, out); 1267 USTPUTC(v, out); 1268 return out; 1269 } 1270 1271 1272 /* 1273 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 1274 * is not NULL, read a here document. In the latter case, eofmark is the 1275 * word which marks the end of the document and striptabs is true if 1276 * leading tabs should be stripped from the document. The argument firstc 1277 * is the first character of the input token or document. 1278 * 1279 * Because C does not have internal subroutines, I have simulated them 1280 * using goto's to implement the subroutine linkage. The following macros 1281 * will run code that appears at the end of readtoken1. 1282 */ 1283 1284 #define CHECKEND() {goto checkend; checkend_return:;} 1285 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 1286 #define PARSESUB() {goto parsesub; parsesub_return:;} 1287 #define PARSEARITH() {goto parsearith; parsearith_return:;} 1288 1289 static int 1290 readtoken1(int firstc, char const *initialsyntax, char *eofmark, int striptabs) 1291 { 1292 int c = firstc; 1293 char * volatile out; 1294 int len; 1295 char line[EOFMARKLEN + 1]; 1296 struct nodelist *bqlist; 1297 volatile int quotef; 1298 int newvarnest; 1299 int level; 1300 int synentry; 1301 struct tokenstate state_static[MAXNEST_STATIC]; 1302 int maxnest = MAXNEST_STATIC; 1303 struct tokenstate *state = state_static; 1304 int sqiscstyle = 0; 1305 1306 startlinno = plinno; 1307 quotef = 0; 1308 bqlist = NULL; 1309 newvarnest = 0; 1310 level = 0; 1311 state[level].syntax = initialsyntax; 1312 state[level].parenlevel = 0; 1313 state[level].category = TSTATE_TOP; 1314 1315 STARTSTACKSTR(out); 1316 loop: { /* for each line, until end of word */ 1317 CHECKEND(); /* set c to PEOF if at end of here document */ 1318 for (;;) { /* until end of line or end of word */ 1319 CHECKSTRSPACE(4, out); /* permit 4 calls to USTPUTC */ 1320 1321 synentry = state[level].syntax[c]; 1322 1323 switch(synentry) { 1324 case CNL: /* '\n' */ 1325 if (state[level].syntax == BASESYNTAX) 1326 goto endword; /* exit outer loop */ 1327 USTPUTC(c, out); 1328 plinno++; 1329 if (doprompt) 1330 setprompt(2); 1331 else 1332 setprompt(0); 1333 c = pgetc(); 1334 goto loop; /* continue outer loop */ 1335 case CSBACK: 1336 if (sqiscstyle) { 1337 out = readcstyleesc(out); 1338 break; 1339 } 1340 /* FALLTHROUGH */ 1341 case CWORD: 1342 USTPUTC(c, out); 1343 break; 1344 case CCTL: 1345 if (eofmark == NULL || initialsyntax != SQSYNTAX) 1346 USTPUTC(CTLESC, out); 1347 USTPUTC(c, out); 1348 break; 1349 case CBACK: /* backslash */ 1350 c = pgetc(); 1351 if (c == PEOF) { 1352 USTPUTC('\\', out); 1353 pungetc(); 1354 } else if (c == '\n') { 1355 plinno++; 1356 if (doprompt) 1357 setprompt(2); 1358 else 1359 setprompt(0); 1360 } else { 1361 if (state[level].syntax == DQSYNTAX && 1362 c != '\\' && c != '`' && c != '$' && 1363 (c != '"' || (eofmark != NULL && 1364 newvarnest == 0)) && 1365 (c != '}' || state[level].category != TSTATE_VAR_OLD)) 1366 USTPUTC('\\', out); 1367 if ((eofmark == NULL || 1368 newvarnest > 0) && 1369 state[level].syntax == BASESYNTAX) 1370 USTPUTC(CTLQUOTEMARK, out); 1371 if (SQSYNTAX[c] == CCTL) 1372 USTPUTC(CTLESC, out); 1373 USTPUTC(c, out); 1374 if ((eofmark == NULL || 1375 newvarnest > 0) && 1376 state[level].syntax == BASESYNTAX && 1377 state[level].category == TSTATE_VAR_OLD) 1378 USTPUTC(CTLQUOTEEND, out); 1379 quotef++; 1380 } 1381 break; 1382 case CSQUOTE: 1383 USTPUTC(CTLQUOTEMARK, out); 1384 state[level].syntax = SQSYNTAX; 1385 sqiscstyle = 0; 1386 break; 1387 case CDQUOTE: 1388 USTPUTC(CTLQUOTEMARK, out); 1389 state[level].syntax = DQSYNTAX; 1390 break; 1391 case CENDQUOTE: 1392 if (eofmark != NULL && newvarnest == 0) 1393 USTPUTC(c, out); 1394 else { 1395 if (state[level].category == TSTATE_VAR_OLD) 1396 USTPUTC(CTLQUOTEEND, out); 1397 state[level].syntax = BASESYNTAX; 1398 quotef++; 1399 } 1400 break; 1401 case CVAR: /* '$' */ 1402 PARSESUB(); /* parse substitution */ 1403 break; 1404 case CENDVAR: /* '}' */ 1405 if (level > 0 && 1406 ((state[level].category == TSTATE_VAR_OLD && 1407 state[level].syntax == 1408 state[level - 1].syntax) || 1409 (state[level].category == TSTATE_VAR_NEW && 1410 state[level].syntax == BASESYNTAX))) { 1411 if (state[level].category == TSTATE_VAR_NEW) 1412 newvarnest--; 1413 level--; 1414 USTPUTC(CTLENDVAR, out); 1415 } else { 1416 USTPUTC(c, out); 1417 } 1418 break; 1419 case CLP: /* '(' in arithmetic */ 1420 state[level].parenlevel++; 1421 USTPUTC(c, out); 1422 break; 1423 case CRP: /* ')' in arithmetic */ 1424 if (state[level].parenlevel > 0) { 1425 USTPUTC(c, out); 1426 --state[level].parenlevel; 1427 } else { 1428 if (pgetc() == ')') { 1429 if (level > 0 && 1430 state[level].category == TSTATE_ARITH) { 1431 level--; 1432 USTPUTC(CTLENDARI, out); 1433 } else 1434 USTPUTC(')', out); 1435 } else { 1436 /* 1437 * unbalanced parens 1438 * (don't 2nd guess - no error) 1439 */ 1440 pungetc(); 1441 USTPUTC(')', out); 1442 } 1443 } 1444 break; 1445 case CBQUOTE: /* '`' */ 1446 out = parsebackq(out, &bqlist, 1, 1447 state[level].syntax == DQSYNTAX && 1448 (eofmark == NULL || newvarnest > 0), 1449 state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX); 1450 break; 1451 case CEOF: 1452 goto endword; /* exit outer loop */ 1453 case CIGN: 1454 break; 1455 default: 1456 if (level == 0) 1457 goto endword; /* exit outer loop */ 1458 USTPUTC(c, out); 1459 } 1460 c = pgetc_macro(); 1461 } 1462 } 1463 endword: 1464 if (state[level].syntax == ARISYNTAX) 1465 synerror("Missing '))'"); 1466 if (state[level].syntax != BASESYNTAX && eofmark == NULL) 1467 synerror("Unterminated quoted string"); 1468 if (state[level].category == TSTATE_VAR_OLD || 1469 state[level].category == TSTATE_VAR_NEW) { 1470 startlinno = plinno; 1471 synerror("Missing '}'"); 1472 } 1473 if (state != state_static) 1474 parser_temp_free_upto(state); 1475 USTPUTC('\0', out); 1476 len = out - stackblock(); 1477 out = stackblock(); 1478 if (eofmark == NULL) { 1479 if ((c == '>' || c == '<') 1480 && quotef == 0 1481 && len <= 2 1482 && (*out == '\0' || is_digit(*out))) { 1483 PARSEREDIR(); 1484 return lasttoken = TREDIR; 1485 } else { 1486 pungetc(); 1487 } 1488 } 1489 quoteflag = quotef; 1490 backquotelist = bqlist; 1491 grabstackblock(len); 1492 wordtext = out; 1493 return lasttoken = TWORD; 1494 /* end of readtoken routine */ 1495 1496 1497 /* 1498 * Check to see whether we are at the end of the here document. When this 1499 * is called, c is set to the first character of the next input line. If 1500 * we are at the end of the here document, this routine sets the c to PEOF. 1501 */ 1502 1503 checkend: { 1504 if (eofmark) { 1505 if (striptabs) { 1506 while (c == '\t') 1507 c = pgetc(); 1508 } 1509 if (c == *eofmark) { 1510 if (pfgets(line, sizeof line) != NULL) { 1511 char *p, *q; 1512 1513 p = line; 1514 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1515 if ((*p == '\0' || *p == '\n') && *q == '\0') { 1516 c = PEOF; 1517 if (*p == '\n') { 1518 plinno++; 1519 needprompt = doprompt; 1520 } 1521 } else { 1522 pushstring(line, strlen(line), NULL); 1523 } 1524 } 1525 } 1526 } 1527 goto checkend_return; 1528 } 1529 1530 1531 /* 1532 * Parse a redirection operator. The variable "out" points to a string 1533 * specifying the fd to be redirected. The variable "c" contains the 1534 * first character of the redirection operator. 1535 */ 1536 1537 parseredir: { 1538 char fd = *out; 1539 union node *np; 1540 1541 np = (union node *)stalloc(sizeof (struct nfile)); 1542 if (c == '>') { 1543 np->nfile.fd = 1; 1544 c = pgetc(); 1545 if (c == '>') 1546 np->type = NAPPEND; 1547 else if (c == '&') 1548 np->type = NTOFD; 1549 else if (c == '|') 1550 np->type = NCLOBBER; 1551 else { 1552 np->type = NTO; 1553 pungetc(); 1554 } 1555 } else { /* c == '<' */ 1556 np->nfile.fd = 0; 1557 c = pgetc(); 1558 if (c == '<') { 1559 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1560 np = (union node *)stalloc(sizeof (struct nhere)); 1561 np->nfile.fd = 0; 1562 } 1563 np->type = NHERE; 1564 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1565 heredoc->here = np; 1566 if ((c = pgetc()) == '-') { 1567 heredoc->striptabs = 1; 1568 } else { 1569 heredoc->striptabs = 0; 1570 pungetc(); 1571 } 1572 } else if (c == '&') 1573 np->type = NFROMFD; 1574 else if (c == '>') 1575 np->type = NFROMTO; 1576 else { 1577 np->type = NFROM; 1578 pungetc(); 1579 } 1580 } 1581 if (fd != '\0') 1582 np->nfile.fd = digit_val(fd); 1583 redirnode = np; 1584 goto parseredir_return; 1585 } 1586 1587 1588 /* 1589 * Parse a substitution. At this point, we have read the dollar sign 1590 * and nothing else. 1591 */ 1592 1593 parsesub: { 1594 char buf[10]; 1595 int subtype; 1596 int typeloc; 1597 int flags; 1598 char *p; 1599 static const char types[] = "}-+?="; 1600 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1601 int linno; 1602 int length; 1603 int c1; 1604 1605 c = pgetc(); 1606 if (c == '(') { /* $(command) or $((arith)) */ 1607 if (pgetc() == '(') { 1608 PARSEARITH(); 1609 } else { 1610 pungetc(); 1611 out = parsebackq(out, &bqlist, 0, 1612 state[level].syntax == DQSYNTAX && 1613 (eofmark == NULL || newvarnest > 0), 1614 state[level].syntax == DQSYNTAX || 1615 state[level].syntax == ARISYNTAX); 1616 } 1617 } else if (c == '{' || is_name(c) || is_special(c)) { 1618 USTPUTC(CTLVAR, out); 1619 typeloc = out - stackblock(); 1620 USTPUTC(VSNORMAL, out); 1621 subtype = VSNORMAL; 1622 flags = 0; 1623 if (c == '{') { 1624 bracketed_name = 1; 1625 c = pgetc(); 1626 subtype = 0; 1627 } 1628 varname: 1629 if (!is_eof(c) && is_name(c)) { 1630 length = 0; 1631 do { 1632 STPUTC(c, out); 1633 c = pgetc(); 1634 length++; 1635 } while (!is_eof(c) && is_in_name(c)); 1636 if (length == 6 && 1637 strncmp(out - length, "LINENO", length) == 0) { 1638 /* Replace the variable name with the 1639 * current line number. */ 1640 linno = plinno; 1641 if (funclinno != 0) 1642 linno -= funclinno - 1; 1643 snprintf(buf, sizeof(buf), "%d", linno); 1644 STADJUST(-6, out); 1645 STPUTS(buf, out); 1646 flags |= VSLINENO; 1647 } 1648 } else if (is_digit(c)) { 1649 if (bracketed_name) { 1650 do { 1651 STPUTC(c, out); 1652 c = pgetc(); 1653 } while (is_digit(c)); 1654 } else { 1655 STPUTC(c, out); 1656 c = pgetc(); 1657 } 1658 } else if (is_special(c)) { 1659 c1 = c; 1660 c = pgetc(); 1661 if (subtype == 0 && c1 == '#') { 1662 subtype = VSLENGTH; 1663 if (strchr(types, c) == NULL && c != ':' && 1664 c != '#' && c != '%') 1665 goto varname; 1666 c1 = c; 1667 c = pgetc(); 1668 if (c1 != '}' && c == '}') { 1669 pungetc(); 1670 c = c1; 1671 goto varname; 1672 } 1673 pungetc(); 1674 c = c1; 1675 c1 = '#'; 1676 subtype = 0; 1677 } 1678 USTPUTC(c1, out); 1679 } else { 1680 subtype = VSERROR; 1681 if (c == '}') 1682 pungetc(); 1683 else if (c == '\n' || c == PEOF) 1684 synerror("Unexpected end of line in substitution"); 1685 else 1686 USTPUTC(c, out); 1687 } 1688 if (subtype == 0) { 1689 switch (c) { 1690 case ':': 1691 flags |= VSNUL; 1692 c = pgetc(); 1693 /*FALLTHROUGH*/ 1694 default: 1695 p = strchr(types, c); 1696 if (p == NULL) { 1697 if (c == '\n' || c == PEOF) 1698 synerror("Unexpected end of line in substitution"); 1699 if (flags == VSNUL) 1700 STPUTC(':', out); 1701 STPUTC(c, out); 1702 subtype = VSERROR; 1703 } else 1704 subtype = p - types + VSNORMAL; 1705 break; 1706 case '%': 1707 case '#': 1708 { 1709 int cc = c; 1710 subtype = c == '#' ? VSTRIMLEFT : 1711 VSTRIMRIGHT; 1712 c = pgetc(); 1713 if (c == cc) 1714 subtype++; 1715 else 1716 pungetc(); 1717 break; 1718 } 1719 } 1720 } else if (subtype != VSERROR) { 1721 if (subtype == VSLENGTH && c != '}') 1722 subtype = VSERROR; 1723 pungetc(); 1724 } 1725 STPUTC('=', out); 1726 if (state[level].syntax == DQSYNTAX || 1727 state[level].syntax == ARISYNTAX) 1728 flags |= VSQUOTE; 1729 *(stackblock() + typeloc) = subtype | flags; 1730 if (subtype != VSNORMAL) { 1731 if (level + 1 >= maxnest) { 1732 maxnest *= 2; 1733 if (state == state_static) { 1734 state = parser_temp_alloc( 1735 maxnest * sizeof(*state)); 1736 memcpy(state, state_static, 1737 MAXNEST_STATIC * sizeof(*state)); 1738 } else 1739 state = parser_temp_realloc(state, 1740 maxnest * sizeof(*state)); 1741 } 1742 level++; 1743 state[level].parenlevel = 0; 1744 if (subtype == VSMINUS || subtype == VSPLUS || 1745 subtype == VSQUESTION || subtype == VSASSIGN) { 1746 /* 1747 * For operators that were in the Bourne shell, 1748 * inherit the double-quote state. 1749 */ 1750 state[level].syntax = state[level - 1].syntax; 1751 state[level].category = TSTATE_VAR_OLD; 1752 } else { 1753 /* 1754 * The other operators take a pattern, 1755 * so go to BASESYNTAX. 1756 * Also, ' and " are now special, even 1757 * in here documents. 1758 */ 1759 state[level].syntax = BASESYNTAX; 1760 state[level].category = TSTATE_VAR_NEW; 1761 newvarnest++; 1762 } 1763 } 1764 } else if (c == '\'' && state[level].syntax == BASESYNTAX) { 1765 /* $'cstylequotes' */ 1766 USTPUTC(CTLQUOTEMARK, out); 1767 state[level].syntax = SQSYNTAX; 1768 sqiscstyle = 1; 1769 } else { 1770 USTPUTC('$', out); 1771 pungetc(); 1772 } 1773 goto parsesub_return; 1774 } 1775 1776 1777 /* 1778 * Parse an arithmetic expansion (indicate start of one and set state) 1779 */ 1780 parsearith: { 1781 1782 if (level + 1 >= maxnest) { 1783 maxnest *= 2; 1784 if (state == state_static) { 1785 state = parser_temp_alloc( 1786 maxnest * sizeof(*state)); 1787 memcpy(state, state_static, 1788 MAXNEST_STATIC * sizeof(*state)); 1789 } else 1790 state = parser_temp_realloc(state, 1791 maxnest * sizeof(*state)); 1792 } 1793 level++; 1794 state[level].syntax = ARISYNTAX; 1795 state[level].parenlevel = 0; 1796 state[level].category = TSTATE_ARITH; 1797 USTPUTC(CTLARI, out); 1798 if (state[level - 1].syntax == DQSYNTAX) 1799 USTPUTC('"',out); 1800 else 1801 USTPUTC(' ',out); 1802 goto parsearith_return; 1803 } 1804 1805 } /* end of readtoken */ 1806 1807 1808 1809 #ifdef mkinit 1810 RESET { 1811 tokpushback = 0; 1812 checkkwd = 0; 1813 } 1814 #endif 1815 1816 /* 1817 * Returns true if the text contains nothing to expand (no dollar signs 1818 * or backquotes). 1819 */ 1820 1821 static int 1822 noexpand(char *text) 1823 { 1824 char *p; 1825 char c; 1826 1827 p = text; 1828 while ((c = *p++) != '\0') { 1829 if ( c == CTLQUOTEMARK) 1830 continue; 1831 if (c == CTLESC) 1832 p++; 1833 else if (BASESYNTAX[(int)c] == CCTL) 1834 return 0; 1835 } 1836 return 1; 1837 } 1838 1839 1840 /* 1841 * Return true if the argument is a legal variable name (a letter or 1842 * underscore followed by zero or more letters, underscores, and digits). 1843 */ 1844 1845 int 1846 goodname(const char *name) 1847 { 1848 const char *p; 1849 1850 p = name; 1851 if (! is_name(*p)) 1852 return 0; 1853 while (*++p) { 1854 if (! is_in_name(*p)) 1855 return 0; 1856 } 1857 return 1; 1858 } 1859 1860 1861 /* 1862 * Called when an unexpected token is read during the parse. The argument 1863 * is the token that is expected, or -1 if more than one type of token can 1864 * occur at this point. 1865 */ 1866 1867 static void 1868 synexpect(int token) 1869 { 1870 char msg[64]; 1871 1872 if (token >= 0) { 1873 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1874 tokname[lasttoken], tokname[token]); 1875 } else { 1876 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1877 } 1878 synerror(msg); 1879 } 1880 1881 1882 static void 1883 synerror(const char *msg) 1884 { 1885 if (commandname) 1886 outfmt(out2, "%s: %d: ", commandname, startlinno); 1887 outfmt(out2, "Syntax error: %s\n", msg); 1888 error(NULL); 1889 } 1890 1891 static void 1892 setprompt(int which) 1893 { 1894 whichprompt = which; 1895 1896 #ifndef NO_HISTORY 1897 if (!el) 1898 #endif 1899 { 1900 out2str(getprompt(NULL)); 1901 flushout(out2); 1902 } 1903 } 1904 1905 /* 1906 * called by editline -- any expansions to the prompt 1907 * should be added here. 1908 */ 1909 const char * 1910 getprompt(void *unused __unused) 1911 { 1912 static char ps[PROMPTLEN]; 1913 const char *fmt; 1914 const char *pwd; 1915 int i, trim; 1916 1917 /* 1918 * Select prompt format. 1919 */ 1920 switch (whichprompt) { 1921 case 0: 1922 fmt = ""; 1923 break; 1924 case 1: 1925 fmt = ps1val(); 1926 break; 1927 case 2: 1928 fmt = ps2val(); 1929 break; 1930 default: 1931 return "??"; 1932 } 1933 1934 /* 1935 * Format prompt string. 1936 */ 1937 for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++) 1938 if (*fmt == '\\') 1939 switch (*++fmt) { 1940 1941 /* 1942 * Hostname. 1943 * 1944 * \h specifies just the local hostname, 1945 * \H specifies fully-qualified hostname. 1946 */ 1947 case 'h': 1948 case 'H': 1949 ps[i] = '\0'; 1950 gethostname(&ps[i], PROMPTLEN - i); 1951 /* Skip to end of hostname. */ 1952 trim = (*fmt == 'h') ? '.' : '\0'; 1953 while ((ps[i+1] != '\0') && (ps[i+1] != trim)) 1954 i++; 1955 break; 1956 1957 /* 1958 * Working directory. 1959 * 1960 * \W specifies just the final component, 1961 * \w specifies the entire path. 1962 */ 1963 case 'W': 1964 case 'w': 1965 pwd = lookupvar("PWD"); 1966 if (pwd == NULL) 1967 pwd = "?"; 1968 if (*fmt == 'W' && 1969 *pwd == '/' && pwd[1] != '\0') 1970 strlcpy(&ps[i], strrchr(pwd, '/') + 1, 1971 PROMPTLEN - i); 1972 else 1973 strlcpy(&ps[i], pwd, PROMPTLEN - i); 1974 /* Skip to end of path. */ 1975 while (ps[i + 1] != '\0') 1976 i++; 1977 break; 1978 1979 /* 1980 * Superuser status. 1981 * 1982 * '$' for normal users, '#' for root. 1983 */ 1984 case '$': 1985 ps[i] = (geteuid() != 0) ? '$' : '#'; 1986 break; 1987 1988 /* 1989 * A literal \. 1990 */ 1991 case '\\': 1992 ps[i] = '\\'; 1993 break; 1994 1995 /* 1996 * Emit unrecognized formats verbatim. 1997 */ 1998 default: 1999 ps[i++] = '\\'; 2000 ps[i] = *fmt; 2001 break; 2002 } 2003 else 2004 ps[i] = *fmt; 2005 ps[i] = '\0'; 2006 return (ps); 2007 } 2008